feat(compose): internal db network and api healthcheck

Add the internal-only legacyhub_db network (same name the teamhub
federation overlay already uses) plus an edge network to the base
compose. Dev keeps host-published ports via edge; the prod overlay
pins data services to legacyhub_db only, closing the module-contract
gap (DB/broker on internal networks). Add a curl liveness healthcheck
for the api container against /api/v1/health.

Verified: docker compose config for dev, prod and prod+teamhub
stacks; per-service network/port/healthcheck matrix inspected via
config --format json.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

RUNBOOK.md

@@ -27,7 +27,8 @@ cd frontend && cp .env.example .env && npm install && npm run dev
 ## Production deploy
 
 Production overlay enables OpenSearch security plugin, removes default ports,
-forces externally-supplied credentials, and disables debug routes.
+pins data services to the internal-only `legacyhub_db` network (no host access,
+no egress), forces externally-supplied credentials, and disables debug routes.
 
 ```bash
 # 1. Ensure secrets exist
@@ -142,8 +143,11 @@ Two mechanisms layered together:
 ## Network model & firewall (federation)
 
 - **Standalone dev** (`docker-compose.yml`): runs the full stack and publishes
-  ports on localhost for convenience. Direct `:8050` access bypasses the
+  ports on localhost for convenience. Data services sit on the internal-only
-  gateway and must never be exposed beyond the workstation.
+  `legacyhub_db` network plus `edge` (so the published ports keep working);
+  the prod overlay drops the `edge` attachment. The `api` container carries a
+  docker healthcheck against `/api/v1/health`. Direct `:8050` access bypasses
+  the gateway and must never be exposed beyond the workstation.
 - **Federated / prod**: add `docker-compose.teamhub.yml`. It moves every backing
   service onto an internal-only network (no host ports), attaches the `api` to
   the shared `teamhub_net`, stops publishing the api port (the gateway reaches

docker-compose.prod.yml

@@ -10,6 +10,7 @@
 #
 # This overlay narrows the dev-friendly defaults:
 #   - removes published ports from data services (only api stays public);
+#   - pins data services to the internal-only docker network (no egress);
 #   - turns on the OpenSearch security plugin and forces an admin password;
 #   - requires CORS_ALLOWED_ORIGINS to be set (no localhost fallback);
 #   - bumps Java + worker concurrency for real workloads;
@@ -17,6 +18,7 @@
 
 services:
   postgres:
+    networks: !override [legacyhub_db]
     ports: !reset []
     environment:
       POSTGRES_DB: ${POSTGRES_DB}
@@ -26,6 +28,7 @@ services:
 
   minio:
     command: server /data
+    networks: !override [legacyhub_db]
     ports: !reset []
     environment:
       MINIO_ROOT_USER: ${MINIO_ACCESS_KEY:?MINIO_ACCESS_KEY must be set}
@@ -33,6 +36,7 @@ services:
     restart: always
 
   opensearch:
+    networks: !override [legacyhub_db]
     ports: !reset []
     environment:
       - discovery.type=single-node
@@ -44,12 +48,14 @@ services:
     restart: always
 
   qdrant:
+    networks: !override [legacyhub_db]
     ports: !reset []
     environment:
       QDRANT__SERVICE__API_KEY: ${QDRANT_API_KEY:?QDRANT_API_KEY must be set}
     restart: always
 
   redis:
+    networks: !override [legacyhub_db]
     ports: !reset []
     restart: always
 

docker-compose.yml

@@ -37,10 +37,22 @@ x-common-env: &common-env
   CORS_ALLOWED_ORIGINS: ${CORS_ALLOWED_ORIGINS:-http://localhost:5273,http://localhost:4173}
   API_KEY: ${API_KEY:-}
 
+# Module-contract isolation: data services live on the internal-only
+# legacyhub_db network (the same name docker-compose.teamhub.yml uses, so the
+# federation overlay composes without a parallel network). In dev they also
+# join edge so host-published ports keep working; the prod overlay pins them
+# to legacyhub_db only. api/worker need edge for the published API port and
+# HuggingFace model downloads.
+networks:
+  legacyhub_db:
+    internal: true
+  edge: {}
+
 services:
   postgres:
     image: postgres:16-alpine
     restart: unless-stopped
+    networks: [legacyhub_db, edge]
     environment:
       POSTGRES_DB: ${POSTGRES_DB:-legacyhub}
       POSTGRES_USER: ${POSTGRES_USER:-legacyhub}
@@ -58,6 +70,7 @@ services:
   minio:
     image: minio/minio:RELEASE.2024-08-29T01-40-52Z
     restart: unless-stopped
+    networks: [legacyhub_db, edge]
     command: server /data --console-address ":9001"
     environment:
       MINIO_ROOT_USER: ${MINIO_ACCESS_KEY:-legacyhub}
@@ -76,6 +89,7 @@ services:
   opensearch:
     image: opensearchproject/opensearch:2.15.0
     restart: unless-stopped
+    networks: [legacyhub_db, edge]
     environment:
       - discovery.type=single-node
       - bootstrap.memory_lock=true
@@ -103,6 +117,7 @@ services:
   qdrant:
     image: qdrant/qdrant:v1.11.3
     restart: unless-stopped
+    networks: [legacyhub_db, edge]
     ports:
       - "6333:6333"
       - "6334:6334"
@@ -117,6 +132,7 @@ services:
   redis:
     image: redis:7-alpine
     restart: unless-stopped
+    networks: [legacyhub_db, edge]
     ports:
       - "6379:6379"
     volumes:
@@ -133,11 +149,20 @@ services:
       dockerfile: docker/Dockerfile
     image: legacyhub/api:latest
     restart: unless-stopped
+    networks: [legacyhub_db, edge]
     environment:
       <<: *common-env
       APP_HOST: 0.0.0.0
       APP_PORT: 8000
     command: ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8000"]
+    healthcheck:
+      # /api/v1/health is liveness: it answers HTTP 200 even when a dependency
+      # is degraded, so the container is not restart-looped on data-store issues.
+      test: ["CMD", "curl", "-fsS", "http://localhost:8000/api/v1/health"]
+      interval: 30s
+      timeout: 10s
+      retries: 5
+      start_period: 60s
     ports:
       # Host port 8050 avoids the 8000 collision (SalesHUB / old dispatch dev).
       # Container still listens on 8000. Overridden by ${API_HOST_PORT}.
@@ -164,6 +189,7 @@ services:
       dockerfile: docker/Dockerfile
     image: legacyhub/api:latest
     restart: unless-stopped
+    networks: [legacyhub_db, edge]
     environment:
       <<: *common-env
     command: ["celery", "-A", "app.workers.celery_app", "worker", "--loglevel=INFO", "--concurrency=2"]